Professional Documents
Culture Documents
Wednesday,October21,2014
OscillatorDesign:AndrewStevens
OscillatorPerformanceandAnalysis:JordanWeber
OscillatorDesign#42oflike
OurfirstbuildanditworkedonlyafterweignoredthesimulationandaddedbackinC3in
parallelwiththeinductoranddecreasedtheresistanceonouremitteroscillatorycircuitinorder
toincreaseourAvloop.Therewasalsoasnafuwithsomeoneleaving68kOhmresistorsin
100kOhmbox.Weusedtheminsteadofourplannedvalueof100kOhm.
Schematics
Layout
Layoutoftheboard.Inductorsizethedarkleadsfromthetop6voltrailtotheCollectorofthe
FirstBJTisthelegsoftheinductorontheboard.Anadditionalcapwasplacedalongthetopof
therailtonegatetheinductancecreatedbythelong6voltrail.
SimultaneousEquationCalculations
ThisistheequationsIusedtodeterminethevaluesofcapacitorsandinductorIwouldneedto
haveinordertoswingbetweenthesefrequencies.
InputResistanceandCapacitanceCalculations
1.
2.
3.
4.
5.
6.
PickedC
tobe82pF
1
CalculatedX
atlowestfrequencytofindtheworstcasereactanceforX
c1
c1
FoundequationsthatrelatedRetoRinofthecapacitorvoltagedivider
SolvedforRetomakeRinatleasttwiceasbigasX
pickedalargervaluetomakesure
c1
PickC
tomakeA
tobebetween2and4cameto33pF
2
vloop
CalculateC
usingthelowestC
valueandnotusingC
findwhatC
needstobein
total
delta
3
3
ordertomatchcalculatedsimultaneousequationvalues.
PowerCalculations
Thisismywattagecalculationsafterthefact.Theactualvaluethatweobservedwasworse.A
resistorchangeonthebuffercouldfixthisbutitcouldalsocauseclippingofthesinesignaland
noonewantsthat.
SimulationIterativeProcess
1.
2.
3.
4.
5.
FirstgotmycalculatedvaluesplacedintoADS
Theoutputwastohigh
Putinbufferstage
Increasedemitterresistanceontheoscillatortodecreaseoutput
Decreaseemitterresistanceonbufferstagetoincreaseswingofbuffertoreduce
clipping
6. ThenthrowawaysimulationbecauseitwaswrongandIneededtofollowmycalculation
TheValuesthatwepickedweretominimizepowerconsumptioninordertogettheoscillatorto
functionthoughwehadtoreverttomypreviousdesignwithmycalculations.Thiscausedusto
beoverpowerspecandtohaveaoscillationlargerthanspecaswell.
DesignConclusion
Iusedacommonbasecoreformyoscillator.Thisallowedmetohaveseverallocationstoput
capsinparallelwithmytankcircuitallowingforeaseofplacementofthevaractorfortuning.I
wantedtouseanArmstrongoscillatorbutdecidedagainstitduetoconstraintsontime.The
varactorplacementwithdiodesplacedbacktobackistoinsureaconstantcapacitancevalue
whiletheoscillatorisinoperation.
ThelayoutwhenitcametotheEmittertoCollectorwithintheOscillatorcoreisonly2zeroohm
resistorsfromeachother.Ithoughtaboutdoingitinsuchamannerwherenozeroohmswould
beneededbutfoundthatitwouldhavebeentoodifficultofabuildtobeworthit.
OscillatorPerformanceandAnalysis
Measurement1:SmokeTestandVoltage/PowerConsumptionChecks
Inordertomakesurethatourcircuitisoperatingcorrectly,thefirstcheckthatweperformed
wasalltheDCvoltagevaluesonouramplifiers.TheDCbiasvoltagesarecrucialtomaking
surethattheamplifiersarenotreversedbiased.IftheDCvoltagevaluesdonotlook
reasonable,chancesarethattheamplifierisnotbehavingcorrectlyandthecircuitwillfail.
Weconnectedourcircuittothe9Vsupplyonthebench.Thenweusedadigitalmultimeter
(DMM)toprobethevoltageatthecollector(V
),base(V
),andemitter(V
)onboththe
c
b
e
commonbaseamplifierandthecommoncollector(buffer)amplifierasaneasychecktoensure
thatnothingwentterriblyduringtheconstructionofourtwostagecircuit.Theresultsare
recordedinTable1.
Table1
Amplifier
CollectorVoltage(V
)
c
BaseVoltage(V
)
b
EmitterVoltage(V
)
e
CommonBase
(Oscillator)
5.99V
2.71V
2.43V
CommonCollector
(Buffer)
5.98V
1.88V
1.70V
Thevoltagesarenotexactlyequivalenttoourcalculatedbiasvoltagesbutthisislargelydueto
theradiofrequencysignalintheoscillatorthatweareanalyzing.Thesignalproducedbythe
oscillatormayinterferewiththeDMMandhinderitsvaluetomeasuretheDCvoltage
accurately.Hence,wearemoreconcernedthatourDCvaluesfollowthegeneralruleV
>V
>V
c
b
e
respectivelyforeachamplifier.
Next,wecheckedourcurrentbeingdrawnfromthe9Vsupplytomakesurewedontgoover
thepowerspecificationofamaximum40mW.Inordertomeasurethesupplycurrentand
voltage,weusedtheDMMtoprobeforthevoltageandcurrentbecauseitismoreaccuratethan
readingthevaluesoffthepowersupply.
I
=5.66mA
s
V=9.07V
Power=I(V)=0.00566(9.07)=51.34mW
Accordingtoourmeasurementsthetotalpowerdrawnfromthe9Vsupplyis51.34mW,whichis
morethan40mW.Wearenotwithinpowerspecifications.
10
Measurement2:OscillationMagnitude
Tomeasuretheoscillationamplitudeweconnectedahighimpedanceprobetothespectrum
analyzerandthe9Vsupplytoourpowerlinesonourcircuitboard.
OurfirstsetofmeasurementswewereinterestedinfindingtheV
andV
valuesonthe
c
e
commonbaseamplifierathighandlowfrequencies.Ourresultsforthesetestsaresupported
byGraphs14andTable2.
Graph1
(Graph1:Voltageoscillationatthecollectorathighfrequency)
Graph2
11
(Graph2:Voltageoscillationattheemitterathighfrequency)
Graph3
(Graph3:Voltageoscillationatthecollectoratlowfrequency)
Graph4
12
(Graph4:Voltageoscillationattheemitteratlowfrequency)
Itisimportanttorealizethattheprobesusedtomeasuretheseoscillationshasa10:1
attenuationsotherealvaluesareshowninTable2.
InordertofindvoltagefromdBmweusetheequation,
2
) 20
P dBm = 10log( (RV * Po
V rms =
Table2
P dBm+20
)
10
10(
* 0.05
CollectorVoltage(V
)
c
EmitterVoltage(V
)
e
HighFrequency
241mV
rms
658mV
rms
LowFrequency
178mV
rms
471mV
rms
(Table2:reflectsthevaluescompensatingfora20dBattenuationprobe)
Next,wemeasuredtheoscillatingmagnitudeatthebufferoutputwithandwithouta50load.
Graphs58showtheoutputwaveforeachscenario.Table3givesthevaluesforeach
condition.
Graph5
13
(Graph5:Oscillationmagnitudeathighfrequencyontheoutputbufferwithnoload.)
Graph6
(Graph6:Oscillationmagnitudeatlowfrequencyontheoutputbufferwithnoload.)
Graph7
14
(Graph7:Oscillationmagnitudeathighfrequencyontheoutputbufferwith50load.)
Graph8
(Graph8:Oscillationmagnitudeatlowfrequencyontheoutputbufferwith50load.)
Remember,weuseda20dBattenuationprobesotherealvalueswithoutthe20dBattenuation
isshowninTable3.
15
Table3
NoLoad
50Load
HighFrequency
331mV
rms
75mV
rms
LowFrequency
338mV
rms
130mV
rms
(Table3:reflectstherealvalues,negatingthe20dBattenuationprobe)
Lastly,wecheckedthefrequencyshiftbetweentheloadedandunloadedconditions,athighand
lowfrequencies.
Table6showsthedifferentfrequenciesunderallconditionstestedpreviously.
Table4
Conditions:
HighFrequency
LowFrequency
NoLoad
97.6MHz
79.5MHz
Loaded
96.5MHz
77.6MHz
Tocalculatethephaseshift,weuse:
frequencyshift=frequency
frequency
Unloaded
Loaded
Forthecaseoflowfrequency,thefrequencyshiftis
1.9MHz.
Underhighfrequencyconditions,thefrequencyshiftis
1.1MHz
.
Loadingthecircuitlowersboththeamplitudeandfrequencyofthesignal.Thisisbecausethe
signalcomesintocontactwithotherelementsintheprobeandspectrumanalyzerwhichcauses
attenuation.
Measurement3:AmplitudeandTuningRange
Tomeasuretheamplitudeandtuningrangeweconnectedtheouroscillatortothespectrum
analyzertomeasurethefrequencyrangeandamplitudeattheoutput.Themeasurementstaken
areillustratedinTable4.
Table4
Frequency(MHz)
Magnitude(dBm)
Magnitude(V
)
rms
96.22
2.68
0.164
95.24
2.75
0.163
16
93.93
2.85
0.161
91.91
2.82
0.162
88.98
3.00
0.158
86.04
3.20
0.155
84.09
3.34
0.152
82.00
3.52
0.149
80.50
3.76
0.145
77.70
3.90
0.143
ObviouslytheV
valueswerenotgivenonthespectrumanalyzer,butweusedthefollowing
rms
equationtocalculatetheV
valuesgiveninTable4.
rms
V rms
10
(P dBm
10 )
* 0.05
Graphs910showtherelationbetweenamplitudeandfrequencyfordifferentvaluesofV
.
rms
Graph9
Graph10
17
Theamplitudefallsasthefrequencygetssmaller.Thishappensbecauseasweswingthe
potentiometercounterclockwisethevaractordiodecapacitorplatesgetfurtherapart,creatinga
largercapacitance.Thiscreatesalowerimpedanceatthecollector(asillustratedinthesmall
signaldiagram).Alowerimpedanceatthecollectorcreatesasmallergain.Hence,lower
frequencyleadstolowergain.
Measurement4:OscillatorHarmonicSpectrum
Tomeasuretheharmonicsinouroscillatorweconnectedourcircuittothespectrumanalyzer.
Then,wesetawidesweeprangefrom1MHz2GHz.ThisspectrumisillustratedinGraph9.
TheharmonicvaluesarerecordedinTable5.
Graph9
18
(Graph9:Showstheharmonicsinourcircuit.)
Table5
Frequency(MHz)
Magnitude(dBm)
79.20
4.20
157.40
9.25
231.30
21.60
309.50
27.30
387.80
35.19
464.98
31.42
543.21
43.02
621.43
36.80
695.30
43.50
773.53
44.00
GiventhevaluesinTable5,weestimatethelargestharmonictobeat157.40MHzwithan
amplitudeof9.25dBm.Thecarrierwavetobeat77.2MHzwithanamplitudeof4.20.We
usedthefollowingformulatogetthelevelofthelargestharmonicindBc:
19
AllvaluesinTable5areroughlydivisiblebythecarrierfrequency79.20MHz.Sothereareno
significant,nonharmonicrelatedspurs.
Measurement5:FrequencyDrift
Totestthefrequencydriftonouroscillatorweconnectedittothespectrumanalyzer,making
surethatwewereabletoseeatleast1kHzvariation.Wewereabletodothisbysettingthe
startandstopfrequencyvaluesonthespectrumanalyzer86kHz88kHz,respectively.Then
totestthedriftwemadesuretheenergywasdissipatedinthecircuitbykeepingitdisconnected
frompowerforatleast10minutes.Thenwereconnectedtheoscillatortopowerandrecorded
thethefrequencyevery10seconds.Wewantedtoseefrequencydriftwithrespecttotime.We
recordedthevaluesinTable6.
Table6
Seconds
Frequency(MHz)
87.960
10
87.963
20
87.963
30
87.963
40
87.960
50
87.957
60
87.954
3minutes
87.902
F requencyDrift(kHz/minute) = initialfrequencyfinallfrequency
TotalTime(seconds)
kHz
ShortT ermF requencyDrift(kHz/minute) = 8796087954
60seconds = 0.1 minute
kHz
LongT ermF requencyDrift(kHz/minute) = 8796087902
60seconds = 0.97 minute
Graph11
20
AswecanseefromGraph11,thecircuitwarmsupforabout10seconds.Then,itlevelsoutin
frequencyforabout20seconds.Next,itbeginstodrifttolowerfrequenciesataboutanaverage
rateof0.10kHz/minute.
Then,wewereinterestedinfrequencychangeswhentheoscillatorwassubjectedtoheat,
specifically80thencooledtoroomtemperature.Werecordedfrequencyvaluesevery10
seconds.OurrecordedvaluesareinTable7.
Table7
Seconds
Frequency(MHz)
87.9434
10
87.8196
20
87.7780
30
87.7000
40
87.6609
50
87.6304
60
87.6109
Thehightemperaturecreatesahigherfrequencyshiftinitiallybutastheoscillatorcoolsback
downtoroomtemperature25thefrequencystabilizesbackaround87.6MHz.
21
Measurement6:PhaseNoise
Forourlasttestwetestedthephasenoiseonourtunedamplifier.Todothisweconnectedthe
oscillatortothehighperformanceE4403Bspectrumanalyzerandletthefrequencystabilize.
Wevariedthespanandresolutionbandwidth(RBW)sothatwecanrecordphasenoise
informationtheinformationrecordedisinTable8.ThesupportingevidenceisGraphs1013.
Graph10
Graph11
22
Graph12
Graph13
Table8
23
P
(f)
n
RBW
P
(f)/Hz(dBc/Hz)
noise
20kHz
74.27dBm
100Hz
94.27
23.59kHz
84.08dBm
10Hz
94.08
200kHZ
69.92dBm
1kHz
99.92
2MHz
3.754dBm
100kHz
53.75
WecalculatedP
usingtheformula:
noise
P noise = P
(f)10log(RBW)
n
Graph14
Graph14showsthephasenoiseatdifferentfrequencies.athighfrequencythegreaterthe
phasenoisebecomes.Wedohaveacoupleofstrangemeasurements.Thisisprobablydueto
humanerrorwhentakingpreliminarymeasurements.Timeconstraintshavekeptusfrom
reexaminingourrecordedvalues.
24
OscillatorPerformanceandAnalysisConclusion
Ouroscillatorisntcompletelyaccordingtospecifications.Weendedupdrawing56.34mWfrom
thepowersupply.Wecouldhaveprobablyamendedthisbynotmistakenlyusing68k
resistorstobiasourcircuitanduse100kresistors.Wecouldalsoincreasetheresistanceon
theemittertoourbuffer.Ourtuningfrequenciesarewithinspecifications,itrangesfrom83MHz
98MHzasseeninourmagnitudeoscillationdata.Asweturnourpotentiometerclockwiseour
oscillatorgoestohigherfrequencies.Weusedtemperaturestablecapacitorsandabufferfor
stability.Ourgainvaluesonouroscillatorarewellabove10dBm.Overall,ourcircuitfunctions
mostlywithinspecifications.